The typical lengths of H-bonds in proteins is: 2.2-2.5 Å (strong), 2.5-3.2 Å (moderate strength) and 3.2-4.0 Å (weak).
However when i tried visualizing H-bonds in Pymol with various pdb files as input, many of the H-bonds were having a very short length (~1.9 Å ).
The command I have used is A=>Polar contacts => within selection.
From what i read up in the Pymol documentation, it does not specifically predict H-bonds and the prediction is not very accurate.
I even tried visualizing the H-bonds of the same structures in Chimera, it shows lesser H-bonds, with short bond lengths.
I believe VMD can be used to count the total H-bonds, but I am not sure if we can visualize them, and whether it is accurate.
I would be really grateful if anyone can suggest alternative protein visualization softwares, which allow accurate prediction of H-bond and visualization based on a pdb input.
try free maestro , i am not sure but pymol can read maestro output as well.
Well, most H-bond I see in MD work are in the range of 1.6 to 2.5. VMD can do both, counting them using "measure hbonds" (plus some TCL scripting) and visualize them by the Drawing Method Hbonds. There, you can also adapt the distance and angle thresholds to your needs.
How do you measure HBonds? The values you cite (The typical lengths of H-bonds in proteins is: 2.2-2.5 Å (strong), 2.5-3.2 Å (moderate strength) and 3.2-4.0 Å (weak).) refer to the heavy-atom heavy-atom distance, and cover a very broad range.
If you measure from the hydrogen to the H-bond acceptor, as I assum Norman Geist is doing, you need to deduct ~1 Å for the donor-hydrogen bond length.
A 1.6 Å distance between the donor and acceptor heavy atoms would represent a rather bad VdW class!
"Polar Contacts" is a rather corse approximation. You can set the cut-off for hydrogen bond detection by adjusting two parameters:
set h_bond_cutoff_center, 3.6
set h_bond_cutoff_edge, 3.2
The hydrogen bond geometric criteria used in PyMOL were designed to emulate those used by DSSP
The example scripts in the page for a better way to identify donor and acceptor atoms in ligand interactions, as frequently ligand bond orders are not properly defined in pdb files. https://pymolwiki.org/index.php/Displaying_Biochemical_Properties#Hydrogen_bonds_and_Polar_Contacts
Ravi Saini Thanks, I will try out Free Maestro software, and see how it predicts H-bonds.
Norman Geist Even pymol allows setting cutoff distances, i am not sure sure about the angle though, I will try out VMD Drawing Methods as well for visualization. Thanks
Annemarie Honegger Norman Geist Ravi Saini
I wished to visualize the differences in the H-bonds formed by the native and wild type amino acid with other residues in a protein (point mutation), and tried using Pymol for this purpose.
I am using MISSENSE 3D for structural analysis of effect of the mutation on the protein. What I observed was:
(i) Upon providing the native predicted structure pdb file as input, the predicted lengths of H-bonds falls within the previously cited range (as opposed to when the same file is given as input in Pymol, where the H-bond lengths are shorter, and do not involve the same residues).
(ii) When the output pdb file (of the mutant structure) obtained from MISSENSE is given as an input in Pymol, the H-bond lengths lie in the typical range, though the residues involved differ from MISSENSE predictions.
When I read a research paper describing how MISSENSE analyzes structures (Reference: Ittisoponpisan, S., Islam, S.A., Khanna, T., Alhuzimi, E., David, A. & Sternberg, M.J.E. (2019) Can Predicted Protein 3D Structures Provide Reliable Insights into whether Missense Variants Are Disease Associated? J. Mol. Biol. 431, 2197-2212. https://doi.org/10.1016/j.jmb.2019.04.009 ), I encountered this line "For the three structural features calculated using a distance (disulfide bonds, hydrogen bonds and salt bridges), 1 Å was added to standard distances to allow for errors in the modeling by SCWRL4 and errors in the predicted structure."
Being a Biochem bachelors student without a detailed understanding of Bioinformatics and Structural Biology, I am unable to understand what it means, and its relevance, but I realize it may explain the differences I am observing in H-bond prediction by Pymol and MISSENSE.
I would be really grateful if you could help me out in understanding this concept.
I recommend you DS Discovery studio. the module where you can analyze ligand-protein is free and you can also obtain 2D and 3D figures. https://discover.3ds.com/discovery-studio-visualizer-download
If you want to make any comparison in regard to hydrogen count, make sure you use exactly the same methods for all entities, and communicate the criteria you used when reporting the results. There simply is not a fixed cut-off between what is and what is not a hydrogen bond - with decreasing distance, the energy decreases steeply as Van-der-Waals repulsion overcomes the bonding energy, and with increasing distance the energy gradually tapers off (See figure 1 in this paper: Article Toward a General Theory of Hydrogen Bonding: The Short, Stro...
). You might want to use more stringent criteria for high resolution or highly energy-minimized structures, while more lenient criteria are used to identify potential hydrogen bonds in poorly resolved or modeled structures. In X-ray structures, you may see non-ideal hydrogen bond distances and angles in amine to carboxylate salt bridges, since alternative hydrogen bonding of the amino group to either of the carboxylate oxygens may result in an averaged nitrogen position (e.g. https://www.researchgate.net/post/What_is_the_maximum_number_of_H-bonds_a_single_N_or_O_atom_can_participate_in) .
You can use Discovery Studio, which is a computational simulation platform for drug discovery and biological macromolecules, with corresponding simulation functional modules.
Annemarie Honegger
I in fact assumed we were talking about the hydrogen-acceptor distance, not the donor-acceptor distance. Thank you for pointing out this source of a misunderstanding. This may also explain differences in various methods or tools. Also, to have this said aswell, the angle for Hbonds in VMD is the donor-acceptor-hydrogen angle, so a perfect linear hydrogen bond has a 0° angle, while the distance is the donor-acceptor distance, typically around 3.
That's the cultural difference between experimentalists and those specialized on computational methods: In typical macromolecular X-ray structure determination, hydrogens are not visible, and at typical resolutions of 2-3 Å, small deviations in donor-hydrogen-acceptor angles are within the experimental error, while large deviations of the angles are frequently are due to averaging of different side chain conformations, therefore taking into account the (inferred) hydrogen positions does not add much to the picture. It's a different story for small molecule and very high resolution structures, as well as for NMR structures. And the dynamics of hydrogen bond distances are a handy way to monitor structural fluctuations during dynamics simulations.
Thank you so much everyone for your suggestions!
Thank you Annemarie Honegger for your insights on this subject, made the concept much more clearer to me.
- Badges
- Science topic